This document discusses deep learning approaches for video object tracking, including using fully convolutional networks to learn discriminative features for localization (conv4-3 vs conv5-3), multi-domain convolutional neural networks, correlation filter based tracking using end-to-end representation learning, recurrent neural networks for tracking, spatially supervised recurrent convolutional neural networks, and challenges in multiple object tracking. It also lists some datasets like YouTube-BB and announces upcoming deep learning courses.